AI Agent Chat
Slack-first engineering assistant for PR intake, CI monitoring, merge gating, and GPT-4o powered check summaries.
See project and architectureApplied AI · Backend Systems · Product Engineering
I build AI-enabled products from ambiguous idea to working system.
I’m Saroj, an engineer based in Spain working across applied AI, backend systems, and product development. I build with Python, FastAPI/Django, LLM and agent workflows, ML integrations, and modern web/mobile technologies.
My focus is simple: turn new technical capability into products people can actually use, evaluate, and trust.
Work
Selected work
Flowtrack
Local-first productivity tracker with AI-powered dashboard behavior and no cloud account dependency.
See projectTenslam Gym
Android-based AI trainer with pose estimation, posture feedback, and rep counting for everyday workouts.
See projectMotion Intelligence Experiments
Ongoing work on converting video into structured human-movement information for products and systems.
See projectArchive
Ask my archive
Ask about my writing, public projects, or background. Answers are generated from information published on this site and may be incomplete. Sources and dates are shown where available.
Experience
Career timeline
2025–Present · Tenslam Vision
Founder & Applied AI / Product Engineer
2023–2025 · Tenslam
Co-Founder & ML / Product Developer
2022–2023 · Box99 / TheHeCapp
AI / Product Contributor
2022 · Demium
Entrepreneur in Residence
2020–2022 · Independent Web & Data Projects
Self-employed / Independent · Spain
2018–2020 · Eventti.net
Backend Development Contributor
2013–2016 · Sky Infosys
Computer Technician
Recognition
Selected external validation
March 2026 · Humans in the Loop ALIA Challenge
3rd place in the ALIA challenge, focused on model optimization, iterative refinement, and human-in-the-loop safe AI workflow design.
Writing
Recent writing
Why I’m Building Motion Intelligence
Why structured movement data matters, and why video is only the starting point.
Running Small LLMs Locally on Limited Hardware
What local model experiments taught me about speed, memory, and practical constraints.
Building a Real-Time AI Gym Trainer
Lessons from turning a mobile camera into a feedback loop for training.